Skip to content

Latest commit

 

History

History
11 lines (8 loc) · 730 Bytes

File metadata and controls

11 lines (8 loc) · 730 Bytes

pyhealth.models.HistAID

Historical Sequential Transformers for AI-augmented Diagnostics (HIST-AID). This model implements a multi-modal architecture designed to replicate the clinical workflow of a radiologist by integrating current visual evidence with longitudinal medical history.

The architecture utilizes a Vision Transformer (ViT) for image feature extraction and a BERT-Base encoder for processing sequential radiology reports. These representations are fused via a Transformer-based fusion layer using cross-modal self-attention to generate context-aware diagnostic predictions.

.. autoclass:: pyhealth.models.hist_aid.HistAID
    :members:
    :undoc-members:
    :show-inheritance: